Development of a Prototype Smart Sensing Device for Detection of Counterfeit Medicines

The definition of counterfeit drugs was established by the World Health Organization in 1992 as one that is deliberately and fraudulently mislabeled with respect to its identity and/or source. It can apply to both branded and generic products, may include products with the correct ingredients or with the wrong ingredients, without active ingredients, with insufficient active ingredients, or with fake packaging.

In 2011, the WHO proposed a new definition of substance, spurious, falsely labelled, falsified, and counterfeit” (SSFFC), considering public health impact. As per WHO 2019a, a counterfeit medicine is deliberately and fraudulently mislabeled concerning its identity, composition or source.

Counterfeit products as per WHO, 2019b

• Products made with inappropriate or inaccurate ingredients
• Products having insufficient or non-existent active components
• Products in counterfeit wrapping
• Drugs whose active components differ from those listed on the packet.
• Relabeling expired medication to increase its shelf life
• Products lacking the manufacturer’s name and address
• Expired products
• Medicines without an expiration date
• Products with an inaccurate amount of the declared active ingredient
• Products with a different number of contaminants.^3,6

As per the USFDA, counterfeit medicines/falsified medicines are described as harmful to health because they are being passed off as authentic, may contain the wrong ingredients, contain too little or no active ingredient at all, or contain other harmful ingredients. As per CDSCO, Counterfeit drugs, which are categorized into three types under the Drug and Cosmetic Act, 1940.

Spurious and Adulterated drugs: Drug formulations that are designed to look like another drug, particularly a well-known brand, to deceive the consumer and capitalize on the success of the original product are known as “spurious” or “imitation” drug products. The product may or may not include active substances.

Adulterated medicines: These include an adulterant or substituted product or are tainted with impurities, making them harmful to health.

Grossly substandard drugs: drugs manufactured by authorized producers that have been shown to have significant defects that lower the medication's egregious carelessness or manufacturing non-compliance with GMPs. The defects may broadly be as under:

• Active ingredient quantities that are less than 70% for products that are thermolabile and less than 5% of the allowed limits for products that are thermostable.
• Tablet/capsules failing in disintegration tests wherever prescribed.
• Liquid preparations that exhibit fungal growth.
• Parenteral preparation that exhibits excessive toxicity, pyrogen/endotoxin test failure, or sterility test.
• Vaccines that exhibit deficiencies in moisture content, toxicity, sterility, or efficacy.
• Any adulterant that makes the product harmful to health is present.

Executed in-process controls, or inappropriate storage or transportation circumstances for medicines. Examples of some such defects are as under:

• Broken or chipped tablets
• Spots, discoloration, or uneven coating are present
• The formulation’s color changes
• Net content varies little
• Weight variation failures in formulations
• Formulations that don’t reach the color test.
• Individual instances of foreign materials being present
• Labelling errors include Redline, Schedule H, Rx, NRx, XRx, and nomenclature errors, color, caution, etc.

2. MATERIALS AND COMPONENTS

Electronic Components

• Arduino Nano
• TCS34725 colour sensor
• White LED
• UV LED
• 220 Ω resistors
• Breadboard
• Jumper wires
• 5 V power supply/USB connection
• Black cardboard box

Sensing Materials

• Whatman filter paper, Grade 1
• Sodium nitroprusside
• Sodium hydroxide
• Bromocresol green
• Selected pharmaceutical samples

For colour reaction, Azithromycin and Calcium lactate were selected. For text detection, Calcium lactate, Paracetamol and Omeprazole were selected.

Software

• Arduino IDE for Arduino Nano programming, TCS34725 testing and LED control
• Python for image processing, OCR, data processing and verification
• OpenCV for preprocessing, segmentation and color analysis
• Tesseract OCR for extraction of printed pharmaceutical information
• Microsoft Excel for reference database creation and comparison

3. METHOD

Arduino Nano-Based Color Sensor and LED Setup

The Arduino Nano was placed on the breadboard and connected to a computer using a USB cable. The Arduino Nano was used to control the white LED and UV LED. The LEDs were programmed to operate alternately, with each LED switched on for 3 seconds followed by an off period. Controlled illumination was used to provide consistent lighting during the sensing experiment.

Component	Connection
White LED, short end (-)	220 Ω → GND
White LED, long end (+)	220 Ω → D9
UV LED, short end (-)	220 Ω → GND
UV LED, long end (+)	220 Ω → D8

Arduino Nano and LED setup.

Arduino Nano–TCS34725 color sensor setup.

Arrangement of Components Inside the Black Cardboard Box

Component	Position	Purpose
TCS34725 colour sensor	Top/center, facing downward	Detects RGB color
Paper-based sensor	Directly below TCS34725	Holds the color reaction
White LED	Beside the sensor, facing paper	Provides controlled illumination
UV LED	Beside the sensor, facing paper	Provides controlled illumination
Arduino Nano	Above the cardboard box	Controls sensor and LEDs
Breadboard	Under Arduino Nano	Holds circuit connections
USB opening	Side wall	Connects Nano to laptop
Black cardboard	Surrounds sensing area	Reduces interference from ambient light

Arrangement of components inside the black cardboard box.

Representative paper-based sensing sequence.

Representative color development for the paper-based sensing reaction.

OCR Workflow

Optical Character Recognition (OCR) was incorporated into the prototype to assist in the identification of medicines from printed labels or packaging. Images of medicine packages were captured using a camera or smartphone under controlled lighting conditions. The images were resized, denoised, adjusted for brightness and contrast, converted to grayscale, thresholder, and cropped to the region containing text. The pre-processed image was then supplied to the OCR system, which converted the detected characters into machine-readable text. The extracted text was compared with the information stored in the reference database.

OCR-Based Authentication of Paracetamol

1. Image Acquisition

The images of tablets/medicine packages were captured using a camera or smartphone under controlled lighting conditions.

2. Image Pre-processing

The quality of the captured images are improved before OCR analysis by the following:

• Resize the image
• Remove noise
• Adjust brightness and contrast
• Convert RGB image to grayscale
• Apply thresholding
• Crop the region containing the text

3. Text Detection Using OCR

The pre-processed image is given to the OCR (Optical Character Recognition) system. OCR detects the characters present on the tablet/package and converts the image-based text into machine-readable text.

4. Text Extraction

The detected text was extracted and compared with the text stored in the database.

OCR- based text detection and authentication of Paracetamol

• The captured image of the paracetamol sample was processed using an Optical Character Recognition (OCR) technique to identify and extract the text printed on the medicine packaging.
• The extracted text was compared with the authentic paracetamol information stored in the previously created database. The system verified the drug name, dosage strength, and other available identifying information.
• If the detected text matched the corresponding database entry, the sample was classified as “Matched”. Any mismatch or unidentified text was considered “Suspected” and could indicate a potentially counterfeit or incorrectly labelled product.

Representative Paracetamol package used for OCR-based identification.

Python code for Paracetamol text recognition

import cv2

import pytesseract

# Tesseract OCR path

pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"

# Load paracetamol image

image = cv2.imread("paracetamol.jpg")

if image is None:

    print("Error: Image not found.")

    exit()

# Resize image

image = cv2.resize(image, None, fx=2, fy=2)

# Convert to grayscale

gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# Remove noise

gray = cv2.GaussianBlur(gray, (3, 3), 0)

# Thresholding

_, thresh = cv2.threshold(

    gray, 0, 255,

    cv2.THRESH_BINARY + cv2.THRESH_OTSU

)

# OCR text extraction

text = pytesseract.image_to_string(thresh, config="--psm 6")

# Display detected text

print("Detected Text:")

print(text)

# Check for Paracetamol

if "paracetamol" in text.lower():

    print("Result: PARACETAMOL DETECTED")

else:

    print("Result: PARACETAMOL NOT DETECTED")

# Display processed image

cv2.imshow("Processed Image", thresh)

cv2.waitKey(0)

cv2.destroyAllWindows()

Result    

OCR- based text detection and authentication of Paracetamol

OCR-Based Authentication of Calcium Lactate

The captured Calcium lactate package image was processed using the same OCR workflow. The extracted text was compared with the corresponding authenticated database entry to support preliminary identification.

Representative Calcium lactate package used for reference comparison.

Python code for Calcium lactate text recognition

import cv2

import pytesseract

# Set Tesseract OCR path

pytesseract.pytesseract.tesseract_cmd = (

    r"C:\Program Files\Tesseract-OCR\tesseract.exe"

)

# Load calcium lactate image

image = cv2.imread("calcium_lactate.jpg")

if image is None:

    print("Error: Image not found.")

    exit()

# Resize image for better OCR accuracy

image = cv2.resize(image, None, fx=2, fy=2)

# Convert image to grayscale

gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# Remove noise

gray = cv2.GaussianBlur(gray, (3, 3), 0)

# Apply thresholding

_, thresh = cv2.threshold(

    gray, 0, 255,

   cv2.THRESH_BINARY+cv2.THRESH_OTSU

)

# Perform OCR

text = pytesseract.image_to_string(

    thresh,

    config="--psm 6"

)

# Display detected text

print("\nDetected Text:")

print(text)

# Check for Calcium Lactate

if "calcium lactate" in text.lower():

    print("Result: CALCIUM LACTATE DETECTED")

else:

    print("Result: CALCIUM LACTATE NOT DETECTED")

# Display processed image

cv2.imshow("Processed Image", thresh)

cv2.waitKey(0)

cv2.destroyAllWindows()

Result

OCR- based text detection and authentication of Calcium Lactate

Representative Omeprazole package used for OCR-based identification.

Python code for Omeprazole text recognition

import cv2

import pytesseract

# Set Tesseract OCR path

pytesseract.pytesseract.tesseract_cmd = (

    r"C:\Program Files\Tesseract-OCR\tesseract.exe"

)

# Load Omeprazole image

image = cv2.imread("omeprazole.jpg")

if image is None:

    print("Error: Image not found.")

    exit()

# Resize image

image = cv2.resize(image, None, fx=2, fy=2)

# Convert to grayscale

gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# Remove noise

gray = cv2.GaussianBlur(gray, (3, 3), 0)

# Apply thresholding

_, thresh = cv2.threshold(

    gray, 0, 255,

    cv2.THRESH_BINARY + cv2.THRESH_OTSU

)

# Perform OCR

text = pytesseract.image_to_string(

    thresh,

    config="--psm 6"

)

# Display detected text

print("\nDetected Text:")

print(text)

# Check for Omeprazole

if "omeprazole" in text.lower():

    print("Result: OMEPRAZOLE DETECTED")

else:

    print("Result: OMEPRAZOLE NOT DETECTED")

# Display processed image

cv2.imshow("Processed Image", thresh)

cv2.waitKey(0)

Result

OCR- based text detection and authentication of Omeprazole

IV.       RESULTS AND DISCUSSION

The availability of counterfeit medicines in the pharmaceutical market poses a significant threat to public health. Counterfeit products may have incorrect drug identity, inadequate active pharmaceutical ingredients, improper strength, or misleading packaging information. Conventional authentication techniques generally depend on sophisticated analytical instruments and laboratory-based procedures, which may be expensive, time-consuming, and unsuitable for rapid field-level screening.

The prototype developed in this project combines a paper-based chemical response with electronic colour measurement. For the tested color reactions, the TCS34725 sensor produced distinct RGB/C readings: the Azithromycin test produced R = 102, G = 46, B = 38 and C = 148, with the detected color recorded as blue, whereas the Calcium lactate test produced R = 25, G = 11, B = 9 and C = 38, with the detected color recorded as red. These observations demonstrate that the prototype can capture measurable color responses from the paper-based sensing zone.

The OCR component provided a second screening parameter by extracting printed pharmaceutical information and comparing it with a reference database. The combination of chemical color response and package-text information can therefore provide two independent parameters for preliminary screening. The present findings support the feasibility of the prototype as a rapid, low-cost preliminary screening system; however, the reported results are preliminary and should not be interpreted as definitive pharmaceutical authentication without confirmatory analytical testing.

V.        CONCLUSION

This project presents the development of a portable, low-cost prototype smart sensing device for preliminary screening of suspected counterfeit medicines. The system combines paper-based colorimetric sensing, a TCS34725 RGB color sensor, controlled white/UV illumination, Arduino Nano interfacing, and OCR-based text recognition. The color response and extracted package information can be compared with authenticated reference data to generate a preliminary matched or suspected result. The prototype may reduce dependence on sophisticated laboratory instruments during initial screening and can be further developed into a smartphone-assisted medicine screening device.

VI.       LIMITATIONS AND FUTURE SCOPE

The present work represents a preliminary prototype. The manuscript reports color-response observations for selected medicine/reagent combinations and OCR-based reference comparisons, but it does not provide a sufficiently large validation dataset, sensitivity/specificity analysis, inter-day reproducibility study, or comparison with a validated reference analytical method. These aspects should be addressed in future studies.

Future development may include testing a larger number of authentic and suspected samples, establishing quantitative color thresholds, evaluating repeatability under different lighting and environmental conditions, expanding the reference database, improving OCR robustness, and integrating the sensing unit with a smartphone-based interface.

VII.     ACKNOWLEDGEMENTS

We express our sincere thanks to the Department of Pharmaceutical Chemistry, College of Pharmacy, Madras Medical College, Chennai-03 for supporting us in carrying out this work.

REFERENCES

1. World Health Organization. Counterfeit medicines: report of a WHO/IMPACT consultation. Geneva: World Health Organization; 2006.
2. World Health Organization. WHO global surveillance and monitoring system for substandard and falsified medical products. Geneva: World Health Organization; 2017. ISBN: 978-92-4-151342-5.
3. World Health Organization. Substandard and falsified medical products: scope, scale and harm [Internet]. Geneva: World Health Organization; 2019 Dec 9 [cited 2026 Aug 22].
4. U.S. Food and Drug Administration. Counterfeit medicine [Internet]. Silver Spring (MD): U.S. Food and Drug Administration; [cited 2026 Aug 22].
5. World Health Organization. Definitions of substandard and falsified (SF) medical products [Internet]. Geneva: World Health Organization; [cited 2026 Aug 22].
6. World Health Organization. Substandard and falsified medical products [Internet]. Geneva: World Health Organization; 2024 Dec 3 [cited 2026 Aug 22].